Random, vaguely correlated facts don't make for causation

Mar 25, 2010 15:14 GMT  ·  By

Journalists exploiting sensationalist headlines that bend the truth aren't anything new and the practice isn't going anywhere anytime soon. At the same time, linking the latest hyped-up trend to, well, anything guarantees a popular article. That's how you get a headline like, "Facebook 'linked to rise in syphilis,'" despite there being very little evidence to support it. The story ran by The Telegraph and plenty of other UK newspapers and tabloids got quite a bit of attention, so much so that Facebook has come out to defend itself.

As the story goes, syphilis cases rose four times in the past year compared to 2008 in Sunderland, Durham and Teesside. What's more, Facebook is particularly popular in the region, more so than in the rest of the UK. Finally, at least part of those who contacted the STD used Facebook to arrange 'meetings.' All of this was coming from Professor Peter Kelly, director of public health in Teesside, so it must have been true, right?

Well, not so much. Cases in Teesside went from ten in 2008 to 30 in 2009, hardly a representative batch. But only some of those used the social network to meet partners anyway. The real problem is that the fact that people in the area use Facebook more than the rest of the country doesn't mean that this was the cause of the rise in cases or even that it's related to it in any way. As Facebook points out, there is a difference between correlation and causation and even the correlation in this case is rather week.

As you can expect, Facebook didn't like the implications and called the whole thing "ridiculous." "While it makes for interesting headlines, the assertions made in newspaper reports that Facebook is responsible for the transmission of STDs are ridiculous, exaggerate the comments made by the professor, and ignore the difference between correlation and causation," a Facebook spokesperson said.